Enhancing stability in autonomous driving: Nonlinear control strategies for low-friction roads

Autonomous vehicles face significant challenges in maintaining both precise tracking and motion stability on low-friction roads due to fluctuating road adhesion. To overcome this, we introduce a novel nonlinear tracking control strategy that leverages real-time estimation of the road adhesion coefficient. First, an innovative recognition algorithm combining a Kalman filter (KF) with an extreme learning machine (ELM) is developed to estimate the road adhesion coefficient accurately. Based on this, we design a nonlinear path tracking control framework that considers actuator saturation. Our control model integrates both 2-degree-of-freedom (2-DoF) and 7-degree-of-freedom (7-DoF) representations through model integration, allowing for consistent control input. Additionally, we derive Hamiltonian functions for cooperative control, establish design conditions based on vehicle dynamics, and propose a nonlinear cooperative control method for achieving simultaneous path tracking and lateral stability. Finally, extensive field tests validate the proposed system, demonstrating that it enhances both path-tracking precision and vehicle lateral stability under challenging low-friction conditions.